Move load_data_for_chart to utils

4 files changed, 41 insertions, 17 deletions

diff --git a/app/admin/routes.py b/app/admin/routes.py
index 5c86fad..efb4f57 100644
--- a/app/admin/routes.py
+++ b/app/admin/routes.py
@@ -18,9 +18,13 @@ from app.admin.forms import (
     AdminSelectDatasetForm,
 )
 from app.models import User, Dataset, Task, Annotation
-from app.utils.datasets import get_name_from_dataset, md5sum, dataset_is_demo
+from app.utils.datasets import (
+    get_name_from_dataset,
+    md5sum,
+    dataset_is_demo,
+    load_data_for_chart,
+)
 from app.utils.tasks import generate_auto_assign_tasks
-from app.main.datasets import load_data_for_chart
 
 
 @bp.route("/manage/tasks", methods=("GET", "POST"))
@@ -289,7 +293,7 @@ def view_annotations_by_dataset(dset_id):
             counter += 1
         anno_clean.append(dict(user=uid, index=ann.cp_index))
 
-    data = load_data_for_chart(dataset.name)
+    data = load_data_for_chart(dataset.name, dataset.md5sum)
     data["annotations"] = anno_clean
     return render_template(
         "admin/annotations_by_dataset.html",
diff --git a/app/main/datasets.py b/app/main/datasets.py
index c06b4bf..7245e51 100644
--- a/app/main/datasets.py
+++ b/app/main/datasets.py
@@ -2,18 +2,11 @@
 
 import os
 import json
+import logging
 
-from flask import current_app
+
+from app.utils.datasets import md5sum
 
 
-def load_data_for_chart(name):
-    dataset_dir = os.path.join(
-        current_app.instance_path, current_app.config["DATASET_DIR"]
-    )
-    target_filename = os.path.join(dataset_dir, name + ".json")
-    with open(target_filename, 'rb') as fid:
-        data = json.load(fid)
-    chart_data = [{"value": x} for x in data['series']['V1']['raw']]
-    return {"chart_data": chart_data}
 
 
diff --git a/app/main/routes.py b/app/main/routes.py
index 60dd6e1..d249c5c 100644
--- a/app/main/routes.py
+++ b/app/main/routes.py
@@ -9,7 +9,7 @@ from app import db
 from app.decorators import login_required
 from app.main import bp
 from app.models import Annotation, Task
-from app.main.datasets import load_data_for_chart
+from app.utils.datasets import load_data_for_chart
 
 RUBRIC = """
 <i>Please mark all the points in the time series where an <b>abrupt change</b> 
@@ -99,7 +99,7 @@ def task(task_id):
     if task.done:
         flash("It's not possible to edit annotations at the moment.")
         return redirect(url_for("main.index"))
-    data = load_data_for_chart(task.dataset.name)
+    data = load_data_for_chart(task.dataset.name, task.dataset.md5sum)
     return render_template(
         "annotate/index.html",
         title="Annotate %s" % task.dataset.name,
diff --git a/app/utils/datasets.py b/app/utils/datasets.py
index 078535f..74785cb 100644
--- a/app/utils/datasets.py
+++ b/app/utils/datasets.py
@@ -33,9 +33,15 @@ Author: Gertjan van den Burg
 
 """
 
-import re
-import json
 import hashlib
+import json
+import logging
+import os
+import re
+
+from flask import current_app
+
+logger = logging.getLogger(__file__)
 
 
 def validate_dataset(filename):
@@ -99,3 +105,24 @@ def md5sum(filename):
             hasher.update(buf)
             buf = fid.read(blocksize)
     return hasher.hexdigest()
+
+
+def load_data_for_chart(name, known_md5):
+    dataset_dir = os.path.join(
+        current_app.instance_path, current_app.config["DATASET_DIR"]
+    )
+    target_filename = os.path.join(dataset_dir, name + ".json")
+    if not md5sum(target_filename) == known_md5:
+        logger.error(
+            """
+        MD5 checksum failed for dataset with name: %s.
+        Found: %s.
+        Expected: %s.
+        """
+            % (name, md5sum(target_filename), known_md5)
+        )
+        return None
+    with open(target_filename, "rb") as fid:
+        data = json.load(fid)
+    chart_data = [{"value": x} for x in data["series"]["V1"]["raw"]]
+    return {"chart_data": chart_data}